kaiju2table and kaiju2krona produce different numbers

[tomberben]

I have a metagenome from a mixed culture bioreactor enriched in Archaea whose reads I classified using both Kraken2 and Kaiju. I then merged the classifications using the kaiju-mergeOutputs script with the default 'lca' option. When I use kaiju2table at the family level, I get 23% 'Candidatus Methanoperedenaceae'. However, if I use kaiju2krona and ktImportText and set the max depth of the krona figure to 8 (family), it shows 36% 'Candidatus Methanoperedenaceae'. Is there an explanation for this difference, or did I make a mistake somewhere?

[pmenzel]

Hi, coud you share the output files of both programs (and nodes.dmp and names.dmp used) and the commands that you used? Then I could try tro reproduce the issue.

[LeeBergstrand]

@pmenzel This might be related to https://github.com/bioinformatics-centre/kaiju/issues/181

If your dropping counts of reads that aren't classified down to the rank value then you are going to get different values than the Krona chart.